TerraMosaic Daily Digest: September 28, 2026
Daily Summary
Slope-hazard studies connect failure mechanics to operational decisions. Coupled particle simulations describe how flow depth, velocity, friction, density and block orientation govern displacement and burial by granular flows, yielding a dimensionless scaling intended to narrow post-disaster search zones. Laboratory studies separately resolve gradation-dependent static liquefaction in spoil slopes and competing failure modes in anchored slopes. These results replace a single stability label with mechanism-specific estimates, while their transfer to field events remains bounded by material characterization and model assumptions.
Regional mapping couples inventories with geological context but exposes different uncertainty sources. A small-sample debris-flow study uses non-additive fuzzy integration in Shimian County, while geology-guided instance segmentation targets spectral confusion in high-resolution landslide imagery. Title-level records describe physics-informed change detection in steep terrain and an integrated field-UAV-numerical workflow for open-pit rock slopes; their validation outcomes are not inferred. Typhoon Gaemi landslides and landslide-dam evolution are likewise retained as important study designs without claims beyond the available bibliographic evidence.
Infrastructure studies focus on hazards that evolve through coupled media and loading histories. Earthquake-landslide interaction is modeled for pile-supported bridge piers; reverse-fault creep and rock creep are considered jointly around tunnels; and seismic shield-tunnel fragility is compared across internal configurations. An offshore-platform analysis further shows that hydrodynamic feedback can change control performance, although its five-record, simplified numerical setting is not operational validation.
Method papers increasingly test where learned representations fail. Earth-observation foundation-model studies examine whether frozen embeddings transfer across wildfire settings and whether retraining is warranted from spectral structure. Distribution-shift calibration, derivative-aware neural operators and physics-informed differentiation audits distinguish pointwise accuracy from decision risk, sensitivity fidelity and discretization consistency. The shared lesson is methodological: a model that fits observed outputs may still fail under new geography, sampling schedules or downstream physical decisions.
Key Trends
The strongest contributions separate physical failure modes, embed geological context in observation and test predictive methods under deployment-relevant shifts.
- Failure is decomposed into transport, liquefaction and support modes: Granular-flow burial, spoil-slope liquefaction and anchored-slope reliability use different state variables and failure mechanisms rather than a shared generic stability score.
- Geological context enters remote mapping explicitly: Landslide segmentation incorporates geological features, while steep-terrain change detection and mining-slope workflows combine physical or structural information with imagery.
- Coupled hazards extend infrastructure analysis: Earthquake shaking interacts with slow slope movement, hydrodynamics or tunnel configuration; creep, seepage and excavation processes are retained as time-dependent controls.
- Transfer is tested against geography, sampling and observables: Wildfire embedding audits, irregular-time-series benchmarks and held-out physical-sensor tests expose failures that random in-domain splits can conceal.
- Uncertainty is linked to decisions rather than reported as a single score: Online certificates, conformal coverage, simulator-aware model selection and decision-relevant error metrics evaluate whether uncertainty changes actions or risk.
Selected Papers
Granular-flow burial, spoil-slope liquefaction, anchored-slope reliability and geology-guided landslide mapping lead this issue. Tunnel, bridge, snow, flood and erosion studies extend the hazard coverage, while selected AI and remote-sensing papers contribute specific tests of transfer, uncertainty, physical consistency and resource-constrained observation.
1. Displacement and burial mechanisms of objects by granular flows: towards implications for post-disaster search and rescue
Core Problem: Post-disaster search zones depend on how granular flows displace and bury objects.
Key Innovation: Coupled material-point and discrete-element simulations identify controls on block transport and propose an impact-Froude scaling across submerged modes; rescue implications remain model-based.
2. Effect of particle gradation on static liquefaction and failure evolution in spoil slopes
Core Problem: Particle-size distribution may change the onset and evolution of static liquefaction in spoil slopes.
Key Innovation: The study directly tests gradation-dependent static liquefaction and failure evolution in spoil material.
3. Reliability analysis of anchored slopes considering multiple failure modes
Core Problem: Anchored slopes can fail through competing mechanisms that single-mode reliability estimates omit.
Key Innovation: A reliability framework evaluates multiple failure modes rather than treating the anchorage system through one prescribed limit state.
4. Debris flow susceptibility assessment based on Choquet fuzzy integral with small sample size: A case study of Shimian County, Ya’an City
Core Problem: Regional debris-flow inventories can be too small for conventional data-hungry models.
Key Innovation: A Choquet fuzzy-integral approach models susceptibility in Shimian County while explicitly targeting small-sample conditions.
5. Spectral-confusion-suppressed landslide recognition in high-resolution remote sensing images using geological-feature enhanced Cascade Mask R-CNN
Core Problem: Spectral similarity between landslides and surrounding terrain causes false detections in high-resolution imagery.
Key Innovation: Geological features augment Cascade Mask R-CNN to suppress spectral confusion in landslide recognition.
6. Seismic response of a pile-supported bridge pier under multi-hazard conditions in slow-moving landslides: a numerical investigation
Core Problem: Bridge foundations may experience earthquake loading while embedded in a creeping landslide.
Key Innovation: A numerical study evaluates the seismic response of a pile-supported bridge pier under combined landslide and earthquake conditions.
7. Physics-Informed Causal Heterogeneous Change Detection for Geological Disaster Assessment in Steep Mountainous Terrain
Core Problem: Title-level focus: Steep terrain creates heterogeneous appearance and process changes that generic change detectors may confuse.
Key Innovation: Title-signalled contribution: The title identifies a causal, physics-informed change-detection framework for geological-disaster assessment; no unobserved benchmark results are inferred. Methods, results and validation could not be assessed from a reliable abstract.
8. Structural rock-slope failure analysis in open-pit mines: an integrated workflow based on field mapping, UAV photogrammetry, and numerical modeling
Core Problem: Title-level focus: Structural controls on mine slopes require consistent field, geometric and mechanical evidence.
Key Innovation: Title-signalled contribution: The title integrates field mapping, UAV photogrammetry and numerical modeling for an open-pit slope; validation details were unavailable. Methods, results and validation could not be assessed from a reliable abstract.
9. Assessment of the evolution trend of landslide damming based on a geoscience database and machine learning hybrid model
Core Problem: Title-level focus: Whether a landslide dam persists or fails depends on interacting geomorphic and hydraulic controls.
Key Innovation: Title-signalled contribution: The title combines a geoscience database with machine learning to assess dam-evolution trends; generalization cannot be assessed without an abstract. Methods, results and validation could not be assessed from a reliable abstract.
10. Geological control and mobility of typhoon gaemi-induced landslides in Zixing City, Hunan Province, China
Core Problem: Title-level focus: Typhoon rainfall can initiate failures whose downstream mobility controls consequence.
Key Innovation: Title-signalled contribution: The title examines geological controls and mobility of Gaemi-induced landslides in Zixing; methods and quantitative evidence remain unavailable. Methods, results and validation could not be assessed from a reliable abstract.
11. Bidirectional feedback between shear displacement and pore-water pressure in rainfall-induced landslides: Insight from a large-scale in situ slope test
Core Problem: Title-level focus: Pore-pressure changes and shear displacement evolve together during rainfall-driven failure.
Key Innovation: Title-signalled contribution: The title identifies a large-scale in-situ test of displacement–pore-pressure feedback; no unobserved result is inferred. Methods, results and validation could not be assessed from a reliable abstract.
12. Laboratory Study of the Dynamics of a Pollutant Within a City Block During Urban Flooding
Core Problem: During urban flooding, pollutants from point releases (e.g., sewer overflows) transported in street runoff can invade city blocks, exposing residents to contamination risks.
Key Innovation: The present laboratory study focuses on the intrusion of pollutants from flooded streets into a porous city block under steady flow conditions. With two openings, pollutant transport is driven by two recirculating mean flow cells and slow turbulent diffusion toward the cell cores, resulting in long filling timescales and a homogeneous equilibrium concentration.
13. Learning Regional Snow Water Equivalent and Snow Height Variations from Sentinel-1 InSAR Acquisitions
Core Problem: Regional snow water equivalent and depth are difficult to retrieve consistently.
Key Innovation: Sentinel-1 interferometric acquisitions are used to learn regional SWE and snow-height variations.
14. AutoCF: An Automated LLM-Assisted Ecosystem for Compound Flood Simulation, Evaluation, and Impact Attribution
Core Problem: Compound-flood studies require linked simulation, evaluation and attribution workflows.
Key Innovation: AutoCF integrates model execution and impact attribution, with automated assistance bounded by the underlying physical models and data.
15. Physics-Informed Neural Networks for Depth-Averaged Avalanche Dynamics
Core Problem: Depth-averaged avalanche equations require efficient solutions across evolving terrain and material states.
Key Innovation: PINNs are formulated for depth-averaged avalanche dynamics; numerical validation defines current applicability.
16. Passive and semi-active nonlinear negative stiffness control of an offshore jacket platform under earthquake-induced hydrodynamic interaction
Core Problem: Hydrodynamic feedback can change structural-control effectiveness during earthquakes.
Key Innovation: Passive and semi-active nonlinear-negative-stiffness devices are compared under five ground motions; the simplified model and idealized switching limit operational inference.
17. Characteristics of extreme snowfall-wind-gust events in Finland (1960-2024): frequency, duration, and intensity
Core Problem: Snowfall and wind gusts must be characterized jointly to represent severe winter conditions.
Key Innovation: Sixty-five years of Finnish data quantify event frequency, duration and intensity for coincident snowfall–gust extremes.
18. A spatially and temporally disaggregated inland flood dataset with flood metrics (2000-2024)
Core Problem: Event-resolved flood metrics are difficult to compare across space and time.
Key Innovation: A spatially and temporally disaggregated dataset covers inland floods from 2000–2024 and exposes reusable event metrics.
19. A morphological regularization-constrained fuzzy C-means clustering algorithm for tunnel rock discontinuity identification
Core Problem: Tunnel discontinuities are difficult to separate when morphology is noisy or incomplete.
Key Innovation: Morphological regularization constrains fuzzy clustering for rock-discontinuity identification.
20. Shear behavior of grouted rock joints under a cyclic normal load
Core Problem: Normal-load cycling alters grouted-joint shear response.
Key Innovation: Experiments quantify cyclic-normal-load effects on grouted rock joints.
21. Evolution model for rock dynamic parameters and fracture characteristics under cyclic loading
Core Problem: Repeated loading changes rock stiffness and fracture structure.
Key Innovation: An evolution model connects dynamic material parameters with fracture characteristics under cyclic loading.
22. Complementary insights from empirical and machine learning approaches for soil erosion susceptibility mapping in semi-arid mountainous regions
Core Problem: Empirical and machine-learning maps may encode different controls and uncertainties.
Key Innovation: Parallel empirical and learned analyses provide complementary susceptibility evidence in a semi-arid mountain region.
23. Evaluating red-bed soft rock as backfill material: Ground deformation at a Tibetan construction site
Core Problem: Using red-bed soft rock as fill can produce site-scale settlement and deformation.
Key Innovation: A Tibetan construction case evaluates backfill behavior and observed ground deformation.
24. Intelligent crack classification and precursory characteristics in the granite shear fracturing process
Core Problem: Crack-type evolution may provide precursory information before granite shear failure.
Key Innovation: Intelligent crack classification is linked to precursor characteristics during controlled shear fracturing.
25. Ensemble Calibration of Tunnel Deformation in Weak Rock Masses
Core Problem: This study evaluates the mechanical behavior of weak and heterogeneous rock masses along the Honaz Tunnel (SW Türkiye) through 2D finite-element simulations.
Key Innovation: To address the extreme prediction scatter of classical empirical equations in weak formations, an ensemble-based calibration framework was developed by averaging the rock mass deformation modulus ({E}_mass) derived from 15 different empirical formulations. The findings highlight the effectiveness of integrating an unweighted ensemble empirical framework with a stratified validation strategy for robust tunnel support design in complex geomechanical environments.
26. An intelligent framework for predicting excavation deformation risk in tunnels incorporating rockmass classification and cloud-model-enhanced modeling
Core Problem: Title-level focus: Rock-mass class and uncertain deformation indicators must be integrated for excavation-risk grading.
Key Innovation: Title-signalled contribution: A cloud-model-enhanced framework combines rock-mass classification with tunnel-deformation risk prediction. Methods, results and validation could not be assessed from a reliable abstract.
27. Interaction of reverse fault creep and rock creep and its effects on tunnels
Core Problem: Title-level focus: Tunnel response depends on simultaneous fault and rock creep.
Key Innovation: Title-signalled contribution: The study models coupled reverse-fault creep and time-dependent rock deformation around tunnels. Methods, results and validation could not be assessed from a reliable abstract.
28. Transverse seismic performance and fragility analysis of shield tunnels considering different internal structural configurations
Core Problem: Title-level focus: Internal structural configurations change transverse shield-tunnel seismic response.
Key Innovation: Title-signalled contribution: Fragility analyses compare shield-tunnel internal configurations under transverse shaking. Methods, results and validation could not be assessed from a reliable abstract.
29. Synergistic Use of SAR, Optical Data, and Water Balance Modeling to Assess the 2024 Extreme Flood Event in Southern Brazil
Core Problem: Title-level focus: Extreme-flood mapping needs consistency between satellite observations and basin water balance.
Key Innovation: Title-signalled contribution: SAR, optical data and water-balance modeling are combined for the 2024 southern Brazil flood. Methods, results and validation could not be assessed from a reliable abstract.
30. Decision tree-based modeling of soil erosion susceptibility in periglacial landforms of Eastern Anatolian Mountains, Türkiye
Core Problem: Title-level focus: Periglacial landforms create distinct erosion controls that regional maps must retain.
Key Innovation: Title-signalled contribution: Decision trees model erosion susceptibility in the mountains of eastern Türkiye. Methods, results and validation could not be assessed from a reliable abstract.
31. Rapid Damage Estimation Through Social Media Content with Geolocation Using Natural Language Processing: A Study of Historical Texas Tropical Storms
Core Problem: Title-level focus: Social-media reports are rapid but spatially noisy indicators of storm damage.
Key Innovation: Title-signalled contribution: Natural-language processing and geolocation are evaluated on historical Texas tropical storms. Methods, results and validation could not be assessed from a reliable abstract.
32. A Novel Resilience Assessment and Recovery Strategy for Multimodal Passenger Transport Networks Under Typhoon Disasters
Core Problem: Title-level focus: Transport-network resilience depends on both disruption propagation and recovery sequencing.
Key Innovation: Title-signalled contribution: A multimodal passenger-network framework assesses resilience and proposes typhoon recovery strategies. Methods, results and validation could not be assessed from a reliable abstract.
33. Prediction framework for excavation-induced deformation of existing tunnels adjacent to foundation pits considering spatial variability of soil: A hybrid deep learning approach
Core Problem: Title-level focus: Soil spatial variability complicates tunnel response beside foundation pits.
Key Innovation: Title-signalled contribution: Hybrid deep learning models excavation-induced tunnel deformation while retaining spatially variable soil conditions. Methods, results and validation could not be assessed from a reliable abstract.
34. Positioning Road-level Flood Sensing for Disruption Assessment based on Uncertainties and Bayesian Networks
Core Problem: Title-level focus: Road sensors must be placed where uncertain flood observations most improve disruption assessment.
Key Innovation: Title-signalled contribution: Bayesian-network disruption models guide uncertainty-aware placement of road-level flood sensing. Methods, results and validation could not be assessed from a reliable abstract.
35. Mechanism of drainage pressure relief and excavation disturbance for tunnel face adjacent to filled karst cavity
Core Problem: Title-level focus: Drainage relief and excavation disturbance interact near filled karst cavities.
Key Innovation: Title-signalled contribution: The study evaluates the coupled mechanisms affecting a tunnel face adjacent to a filled cavity. Methods, results and validation could not be assessed from a reliable abstract.
36. The Earth in One Gaze: Training-Free Active Focus for UHR Remote Sensing Understanding
Core Problem: Multimodal large language models (MLLMs) must balance local detail against scene context when interpreting ultra-high-resolution (UHR) remote sensing (RS) imagery within a limited visual-input budget.
Key Innovation: Our pilot study finds that a frozen MLLM already produces useful question-guided spatial requests, yet crop-based inspection of the selected regions does not consistently improve its answers. Our analyses show that existing MLLMs can guide where to look in UHR images on their own, and that what they can infer from the selected evidence depends on how that evidence is presented.
37. GeoCR: Learning a Generalist Cloud Removal Prior from Heterogeneous Observations
Core Problem: Cloud removal must generalize across heterogeneous sensors and scenes.
Key Innovation: GeoCR learns a general cloud-removal prior from heterogeneous Earth observations.
38. Reuse or Relearn? A Spectral View of Earth Observation Foundation Models
Core Problem: Reusing or retraining Earth embeddings should depend on the downstream spectral structure.
Key Innovation: A spectral analysis compares reuse and relearning of Earth-observation foundation representations.
39. Derivative-Informed Training of Neural Operators On-the-Fly via Sketched Tangent Consistency
Core Problem: Field-value loss alone may not constrain derivatives needed for mechanics.
Key Innovation: Sketched tangent consistency adds derivative information to neural-operator training on the fly.
40. Generative Priors Conditioned on Natural Language for Bayesian Inversion in PDEs
Core Problem: Inverse priors are difficult to specify for heterogeneous field structures.
Key Innovation: Natural-language-conditioned generative priors are evaluated within Bayesian PDE inversion.
41. Distributed Hydrological Modeling in the Feature Space
Core Problem: Distributed hydrological states are expensive to simulate and observe.
Key Innovation: A feature-space formulation learns distributed hydrological dynamics while retaining spatial structure.
42. The limits of exactness: On the failure of automatic differentiation in physics-informed machine learning
Core Problem: Automatic differentiation can be exact computationally yet inconsistent with the intended continuous physics.
Key Innovation: Failure cases expose limits of exactness in physics-informed learning and motivate discretization-aware verification.
43. Theory Guided and Interpretable Neural Operator Design for Partial Differential Equation Learning
Core Problem: Neural-operator design often hides which physical structure produces generalization.
Key Innovation: Theory-guided operator components are separated and interpreted across PDE tasks.
44. A Review of Deep-learning-based Seismic Data Denoising and Its Promising Paradigm Shift to Foundation Models
Core Problem: Denoising can suppress earthquake signals as well as noise.
Key Innovation: A review organizes deep seismic denoising and evaluates the claimed shift toward foundation models.
45. When local gains fail to transfer: Frozen Earth-observation embeddings across wildfires
Core Problem: Local gains in Earth-observation embeddings may fail under geographic or event shift.
Key Innovation: Frozen embedding tests across wildfire settings emphasize cross-region transfer rather than in-domain accuracy.
46. When Should a Satellite Estimate Be Changed? Stress-Testing Neural Corrections for Evapotranspiration
Core Problem: Learned correction can degrade a satellite estimate when deployment conditions change.
Key Innovation: Stress tests identify when neural corrections should be accepted or rejected rather than applied universally.
47. Adapting neural operators for mechanics decisions under changing operating conditions
Core Problem: A surrogate useful for one operating regime may mislead decisions after conditions change.
Key Innovation: The study adapts neural operators for mechanics decisions under changing operating conditions.
48. Physics-Attested Federated Learning: Securing Collaborative Anomaly Detection in Critical Water Infrastructure
Core Problem: Collaborative anomaly detectors can be poisoned or learn physically impossible patterns.
Key Innovation: Federated learning is constrained by water-system physics to support integrity checks across infrastructure operators; direct geohazard deployment is not demonstrated.
49. PACE-FNO: Physics-Aligned Canonical Equivariance for Fourier Neural Operators
Core Problem: Canonical coordinate choices can break physical symmetries in Fourier operators.
Key Innovation: Physics-aligned canonical equivariance constrains Fourier neural operators.
50. China regional 3 km downscaling based on residual Corrective Diffusion model
Core Problem: A fundamental challenge in numerical weather prediction is efficiently producing high-resolution forecasts.
Key Innovation: This study focuses on statistical downscaling, which uses historical data to learn mappings between low- and high-resolution meteorological fields. Specifically, forecasts of radar composite reflectivity reveal that CorrDiff, as a generative model, is capable of capturing fine-grained meteorological details, yielding more physically realistic predictions than deterministic regression-based downscaling models.
51. Multi-source remote sensing approach for monitoring the Caspian Sea shoreline dynamics
Core Problem: This study presents a scalable methodology for the semi-automated extraction of the Caspian Sea shoreline and water surface based on medium-resolution satellite imagery and cloud computing within the Google Earth Engine environment.
Key Innovation: The proposed approach consists of three stages: generation of reference and working shoreline datasets, validation-based selection of the optimal extraction algorithm, and its subsequent scaling to the entire Caspian Sea. The obtained results confirm the effectiveness of the proposed approach as a basis for scalable monitoring of coastal-zone dynamics and demonstrate its potential for geoecological assessments, forecasting morphodynamic processes, and developing adaptation strategies under conditions of continued Caspian Sea level decline.
52. Lava to Leaf: Remote Sensing of Post-Eruption Ecological Succession on the 2018 Kīlauea Lava Flows
Core Problem: In the years following the 2018 eruption of Kīlauea, vegetation has begun colonizing the new landscape.
Key Innovation: This study investigates the environmental factors driving early primary succession on the 2018 Lower East Rift Zone lava flow. These results indicate that proximity to surviving vegetation and broader environmental gradients are more strongly associated with early vegetation recovery than lava morphology or modeled wind exposure.
53. Sentinel-1 SAR Coherence as a Conservative Alert Layer for Tree Cover Loss in Estonia
Core Problem: Monitoring tree cover loss at national scale requires methods that are weather-independent and operationally scalable.
Key Innovation: Here, we evaluate Sentinel-1 SAR interferometric coherence as a conservative alert layer for tree cover loss in Estonia, using multitemporal Airborne Laser Scanning (ALS) data and visually interpreted samples for validation. At pixel level, coherence-based detection achieved User’s Accuracy of about 77-78% and Producer’s Accuracy of about 41% for tree cover loss, indicating that when loss is flagged it is often correct, yet a substantial fraction of true loss pixels remains undetected.
54. Impact of the gravel content on rill erosion and soil losses on colluvial deposit slopes
Core Problem: This study investigated rill morphology and soil loss characteristics during rill erosion on slopes with different gravel contents, aiming to clarify the regulatory mechanism of gravel content on rill erosion and soil loss.
Key Innovation: A runoff experiment with four flow discharges (2, 4, 8, and 12 L min −1) and four gravel contents (0%, 10%, 30%, and 50%) was used to assess the effects of rill morphology on soil losses and concentration. Path analysis shows that rill width-depth ratio (0.656) is the key morphological indicator for characterizing runoff sediment concentration, while the rill cross-sectional shape index (0.859) is the best indicator for soil loss rate.
55. Spatiotemporal changes of glaciers on Yulong Snow Mountain, southeastern Qinghai-Tibet Plateau, from 1987 to 2024
Core Problem: Yulong Snow Mountain (YSM) represents the southern limit of modern temperate glacier distribution in China.
Key Innovation: This study examines glacier evolution over the period 1987-2024 using multisource remote sensing data and mass balance observations of Baishui River Glacier No. Furthermore, it offers critical empirical evidence for understanding the response mechanisms of monsoonal temperate glaciers to climate change, revealing size-dependent, aspect-controlled, and elevation-sensitive patterns of glacier retreat under regional warming.
56. A Sharp Melt-Bearing Lithosphere-Asthenosphere Boundary Beneath the Korean Peninsula: Evidence From Teleseismic Receiver Functions
Core Problem: The lithosphere-asthenosphere boundary (LAB) provides key constraints on how continental lithosphere responds to tectonic processes.
Key Innovation: We use P- and S-receiver functions to investigate crustal and lithospheric structure beneath the southern KP and adjacent volcanic islands. Our results link the present-day lithospheric heterogeneity to the multi-stage tectonic evolution of the eastern Eurasian margin.
57. BridgeSAR: Measuring Bridge-Water Clearance From Space Using SAR Multipath Signatures
Core Problem: Bridge-water clearance controls safe vessel passage and reflects local water-level changes, but direct measurements are unavailable at many bridges.
Key Innovation: We present BridgeSAR, a geometry-guided method that estimates clearance from multipath scattering stripes in synthetic aperture radar (SAR) amplitude images. bridges and compare the results with air-gap sensors or water-level-derived reference clearances.
58. A Semi-Resolved LES-DEM Framework for Turbulent Bedload Transport From Saltation to Sheet-Flow Regimes
Core Problem: Bedload transport widely occurs in natural environments and encompasses flow regimes from saltation to sheet-flow, with markedly different fluid-particle interaction characteristics.
Key Innovation: Here, we propose a semi-resolved LES-DEM framework to overcome such limitations in the conventional volume-averaged CFD-DEM paradigm. The present method is demonstrated to avoid the conventional grid-size limitation and thus allows the fluid field to be resolved on sufficiently fine grids while preserving accurate fluid-particle coupling.
59. Explicit Semantic Transition Modeling for Open-Vocabulary Remote Sensing Change Detection
Core Problem: Title-level focus: The study addresses the focus identified by its title: Explicit Semantic Transition Modeling for Open-Vocabulary Remote Sensing Change Detection.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
60. Inside the Area of Applicability: Feature-Space Reliability Diagnostics Can Miss Failures in a Grassland Biomass Map
Core Problem: Title-level focus: The study addresses the focus identified by its title: Inside the Area of Applicability: Feature-Space Reliability Diagnostics Can Miss Failures in a Grassland Biomass Map.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
61. An Adaptive Frequency-Domain Filter for Improved Phase Fidelity in InSAR Interferograms
Core Problem: Title-level focus: The study addresses the focus identified by its title: An Adaptive Frequency-Domain Filter for Improved Phase Fidelity in InSAR Interferograms.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
62. A Dual-Polarization-Integrated Time-Series InSAR Framework for Arc-Level Parameter Estimation and Network Densification
Core Problem: Title-level focus: The study addresses the focus identified by its title: A Dual-Polarization-Integrated Time-Series InSAR Framework for Arc-Level Parameter Estimation and Network Densification.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
63. Penetration behavior of rock mass under complex geostress conditions: Experimental and numerical study
Core Problem: Title-level focus: The study addresses the focus identified by its title: Penetration behavior of rock mass under complex geostress conditions: Experimental and numerical study.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
64. Quantitative reconstruction of Holocene soil erosion in the Northern Hemisphere reveals increasing human impacts on erosion dynamics
Core Problem: Title-level focus: The study addresses the focus identified by its title: Quantitative reconstruction of Holocene soil erosion in the Northern Hemisphere reveals increasing human impacts on erosion dynamics.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
65. Detection of soil moisture variations beneath an asphalt layer using L-band SAR
Core Problem: Title-level focus: The study addresses the focus identified by its title: Detection of soil moisture variations beneath an asphalt layer using L-band SAR.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
66. InterMine: A foundation model-based multi-task interaction network for remote sensing interpretation of open-pit coal mining activity
Core Problem: Title-level focus: The study addresses the focus identified by its title: InterMine: A foundation model-based multi-task interaction network for remote sensing interpretation of open-pit coal mining activity.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
67. Uncertainty-aware prediction of compacted snow yield strength using TabPFN: Application to snow runway assessment
Core Problem: Title-level focus: The study addresses the focus identified by its title: Uncertainty-aware prediction of compacted snow yield strength using TabPFN: Application to snow runway assessment.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
68. Engineering classification of tunnels in altered strata: A case study from the Luding-Shimian expressway on the Western Margin of the Yangtze Block
Core Problem: Title-level focus: The study addresses the focus identified by its title: Engineering classification of tunnels in altered strata: A case study from the Luding-Shimian expressway on the Western Margin of the Yangtze Block.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
69. Thermo-mechanical response and fracture mechanisms of granite investigated through experiments and 3D FDEM simulations
Core Problem: Title-level focus: The study addresses the focus identified by its title: Thermo-mechanical response and fracture mechanisms of granite investigated through experiments and 3D FDEM simulations.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
70. A Dual-Objective Meta-Learning Physics-Informed Neural Network for multi-condition hydraulic fracture geometry prediction
Core Problem: Title-level focus: The study addresses the focus identified by its title: A Dual-Objective Meta-Learning Physics-Informed Neural Network for multi-condition hydraulic fracture geometry prediction.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
71. A velocity-implicit elastoplastic Lagrangian particle method for granular flow simulations
Core Problem: Title-level focus: The study addresses the focus identified by its title: A velocity-implicit elastoplastic Lagrangian particle method for granular flow simulations.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
72. Optimal design method for S-shaped tunnels crossing above existing shield tunnels based on theoretical solutions of longitudinal deformation
Core Problem: Title-level focus: The study addresses the focus identified by its title: Optimal design method for S-shaped tunnels crossing above existing shield tunnels based on theoretical solutions of longitudinal deformation.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
73. When Does Domain Adaptation Help on Physical Vibration Sensors? A Held-Out-Bearing Study of Neural-Operator and Convolutional Models
Core Problem: Diagnosing rolling-element bearing faults from vibration is a canonical physical-sensing task and a widely used benchmark for domain adaptation under operating-condition shift.
Key Innovation: Accuracies above 99 percent are commonly reported, but under evaluation splits that place the same physical bearing in both training and test. A second dataset, whose held-out units are fault diameters rather than bearings, shows that the same protocol exposes failures that even a target-supervised model cannot avoid.
74. seq2cause: One Autoregressive Backbone, Four Causal Discovery Tasks in Event Sequences
Core Problem: Complex systems such as vehicles, patients, or genomes emit discrete event sequences whose operative question is causal, not predictive: which events cause which other events, and which cause higher-level outcomes such as failures or diseases?
Key Innovation: We present \textsc{Seq2Cause}, a unified framework that resolves all four regimes through a single shared primitive: a pretrained autoregressive model repurposed as an amortized conditional independence testing engine requiring no task-specific retraining. We establish a prediction-causality duality: the model's excess cross-entropy simultaneously bounds causal identification error across all four regimes, so that every improvement in next-token prediction tightens causal guarantees for free.
75. ANaLOG: Anisotropic Native-Latent Operator Guidance for Solving Inverse Problems
Core Problem: Native-latent guidance is a recent paradigm for solving inverse problems with latent diffusion models.
Key Innovation: We propose ANaLOG, a framework for efficient uncertainty-aware guidance with pretrained latent diffusion models. Experiments across five challenging inverse problems show that ANaLOG improves perceptual reconstruction quality over existing methods while preserving efficiency.
76. Model-Agnostic Online Certificate-Driven Calibration for Time Series Forecasting Under Distribution Shift
Core Problem: Forecast calibration can fail after distribution shift.
Key Innovation: Model-agnostic online certificates regulate recalibration without retraining the base forecaster.
77. Uncertainty-Aware Selection of Online Algorithms with Simulator Ensembles
Core Problem: Online algorithms must be selected when the real environment is only partially observed.
Key Innovation: Simulator ensembles propagate selection uncertainty and test robustness to simulator mismatch.
78. Rethinking Cross-Channel Importance in Time-Series Forecasting
Core Problem: Cross-channel modeling is central to multivariate time-series forecasting, yet channels that are statistically related, predictively useful, and actually used by a trained forecaster are often treated as if they defined the same notion of importance.
Key Innovation: Functional interventions further show that strong forecasters use cross-channel information, while their source-reliance rankings agree little with controlled utility or with one another across iTransformer, TimesNet, and a cross-channel TimeMixer. As a constructive consequence, bounded post-hoc support improves a frozen channel-independent forecaster in 12 of 16 dataset-horizon conditions, with a positive aggregate bootstrap interval.
79. Not All Errors Matter: Decision-Relevant Prediction Error Predicts Planning Quality
Core Problem: Prediction errors with equal magnitude can have different planning consequences.
Key Innovation: The work evaluates forecasts by downstream planning quality rather than aggregate error alone.
80. FoundDSR: A Generalizable Foundation Model with Guided 2D Gaussian Splatting for Depth Super-Resolution
Core Problem: We introduce FoundDSR, a generalizable foundation model for robust depth reconstruction across unseen data distributions using RGB-D pairs.
Key Innovation: Furthermore, to mitigate training instability and bias toward dominant sources caused by distribution gaps in large-scale heterogeneous data, we introduce heterogeneous federated learning that allocates each data source to an independent client for local optimization and global aggregation. Extensive zero-shot evaluations on synthetic, real-world, arbitrary-scale, and noisy conditions demonstrate that FoundDSR consistently outperforms existing state-of-the-art approaches, confirming its strong robustness and generalization to unknown scenes.
81. DiffPTS: Rethinking Diffusion ELBO for Probabilistic Time Series Forecasting
Core Problem: Probabilistic time series forecasting requires modeling and predicting complex and time-varying distributions.
Key Innovation: Building on this principled ELBO reformulation, we propose Diff- PTS, a general framework that enables end-to-end optimization of all components within the ELBO. Across multiple benchmarks, DiffPTS consistently outperforms recent models, achieving state-of-the-art performance with an average CRPS/MSE reduction of over 14.53%/16.55% compared to existing diffusion-based methods.
82. TimeES: Probabilistic and Deterministic Time Series Forecasting via Evolutionary Spectra
Core Problem: Real-world time series are inherently non-stationary, with trends, periodic patterns, and uncertainty evolving over time.
Key Innovation: Motivated by Evolutionary Spectra (ES) theory, we propose TimeES, a general framework that enables probabilistic and deterministic forecasting via the evolutionary spectra theory. Based on a simple linear backbone, our proposed TimeES achieves consistent state-of-the-art performance across both deterministic and probabilistic forecasting tasks, with high efficiency and interpretability.
83. CityToolVQA: Tool-Augmented Visual Question Answering for 3D Spatial Cognition in Urban Low-Altitude Environments
Core Problem: CityToolVQA addresses the weak performance of Vision-Language Models (VLMs) on quantitative tasks in urban low-altitude visual question answering.
Key Innovation: We divide the seven tasks into qualitative and quantitative groups: qualitative questions are answered directly by the VLM, whereas quantitative questions are handled by an external visual-geometric toolchain that performs object grounding, segmentation, depth back-projection, and spatial computation. These results indicate that externalizing explicit 3D geometric computation effectively complements the limited ability of RGB-only VLMs to estimate metric distances and object sizes.
84. Adapting Nonstationary Multi-output Gaussian Processes to Bayesian Optimization
Core Problem: Multi-objective Bayesian optimization (MOBO) commonly relies on independent Gaussian processes (GPs) with stationary kernels, limiting its ability to represent nonstationary structure and share information between objectives.
Key Innovation: We introduce MOLRN-BO, which combines a regularized shared-spectral surrogate with objective-specific residuals, prequential mean correction and tempered covariance scaling, and Pareto-local qLogEHVI optimization with periodic global search. These results demonstrate that nonstationary multi-output surrogates can deliver strong and robust MOBO performance when their structure and use are explicitly adapted to the demands of sequential optimization.
85. Rondo: Unsupervised Discovery of Recurring Temporal Structure
Core Problem: Many real-world time series data exhibit structural properties at multiple scales, from short, recurring units to complex sequences composed of these units.
Key Innovation: We introduce Rondo, an unsupervised approach for modeling recurring hierarchical structure in continuous temporal streams. Evaluations on temporal sequences spanning diverse domains and data modalities show that Rondo outperforms existing unsupervised recurrence-discovery baselines, with particularly pronounced advantages in limited-data and continual-stream settings.
86. STAMP: Predicting Out-of-Distribution Generalization without Target Data
Core Problem: Target labels are unavailable when deciding whether a model will generalize.
Key Innovation: STAMP estimates out-of-distribution performance without labelled target data.
87. SIFT: Enhancing Time Series Foundation Models via Semantic Invariance and Structural Fidelity Fine-Tuning
Core Problem: Time Series Foundation Models (TSFMs) have achieved remarkable zero-shot performance through extensive pre-training on massive time series datasets.
Key Innovation: To address these challenges, we propose SIFT, a robust adaptation method that enhances time series foundation models by preserving Semantic Invariance and structural Fidelity throughout the fine-Tuning process. Extensive experiments on representative TSFMs covering 10 real-world datasets demonstrate that SIFT can significantly enhance the performance of TSFMs.
88. Region-Local Copula Evidence Fusion for Heterogeneous Remote Sensing Change Detection
Core Problem: Different regions and sensors provide unequal evidence of change.
Key Innovation: Local copula fusion combines heterogeneous remote-sensing change evidence while preserving regional dependence.
89. Gradient-Guided Decoupled Adaptation for Geospatial Vision-Language Models
Core Problem: General VLMs must adapt without losing spatial grounding.
Key Innovation: Gradient-decoupled adaptation targets geospatial vision-language tasks.
90. Hierarchical Frequency-Domain Compression of Implicit Geometric Representations for Large-Scale Point Clouds
Core Problem: Large-scale point cloud representations of complex geome tries incur prohibitive computational and memory costs, necessitating compressed implicit representations.
Key Innovation: To ad dress this, we propose a unified framework comprising im plicit geometric field representation, hierarchical frequency domain compression, and conditional high-frequency predic tion. Experiments on a complex-boundary point cloud with more than eight mil lion points demonstrate that the proposed method achieves a higher compression ratio than existing point cloud compres sion methods while maintaining comparable reconstruction quality.
91. PolyTopoBench: A Benchmark for Complex Vector Polygon Generation from Remote Sensing Imagery
Core Problem: Raster metrics do not assess vector topology and geometry.
Key Innovation: PolyTopoBench evaluates complex vector-polygon generation from remote-sensing imagery.
92. Efficient Message Passing for Partial Differential Equation Priors
Core Problem: Prior information for real-world physical quantities is most elegantly expressed via partial differential equations (PDEs).
Key Innovation: In this paper, we propose a novel way to solve PDEs using probabilistic inference on a factor graph. We demonstrate our approach on the first-order advection and the second-order semi-linear Fisher-KPP equations, where it achieves predictive accuracy comparable to a standard baseline while providing structured predictive uncertainty.
93. QSCP: Beyond Class-Name Prompts for Query-Guided Semantic Change Parsing
Core Problem: Traditional change detection (CD) identifies changes between bi-temporal remote sensing images, while semantic change detection (SCD) assigns predefined land-cover classes.
Key Innovation: However, mapping all changes may not meet a user's specific needs. On SECOND, QSCP outperforms RCDNet on synonym, sentence, and transition queries and improves end-to-end semantic prediction over evaluated semantic baselines.
94. Simulation-Free Learning of GP-SDEs from Irregular Observations
Core Problem: Gaussian process stochastic differential equations (GP-SDEs) provide a flexible Bayesian model for unknown continuous-time state dynamics with uncertainty quantification, but learning and inference from noisy and irregular observations remain computationally challenging.
Key Innovation: To address this issue, we propose GP-SDE Matching, a simulation-free variational framework for Bayesian GP drift learning and continuous-time state smoothing. Experiments on the stochastic Lorenz-63 system demonstrate substantially improved drift recovery and state reconstruction under irregular observations, while five system identification benchmarks show robust forecasting under increasing observation sparsity and competitive performance against existing latent-SDE and state-space methods.
95. Perturb-and-Solve: Efficient Learned-Operator Conditioning for Latent Diffusion Inverse Problems
Core Problem: Latent diffusion models serve as powerful priors for solving inverse problems in image restoration, such as deblurring, inpainting, and super-resolution.
Key Innovation: To break this bottleneck, we introduce PASEO (Perturb-And-Solve for Efficient Operator conditioning), a method that uses a small (1M parameters) learned network to degrade diffusion model predictions in latent space. Across super-resolution, deblurring, and inpainting on FFHQ and COCO, PASEO achieves strong perceptual quality while running up to 9x faster and using up to 34% less peak memory than the tested baselines, with the same or fewer model evaluations.
96. SMORE: Stability-Promoting Mesh-Agnostic Model Reduction for Time-Dependent PDEs
Core Problem: Reduced models can become unstable when meshes or time regimes change.
Key Innovation: Stability-promoting, mesh-agnostic reduction targets time-dependent PDEs.
97. AevaScenes: An FMCW LiDAR Dataset and Benchmark for Long-Range Perception
Core Problem: Long-range FMCW LiDAR lacks common data for perception evaluation.
Key Innovation: AevaScenes supplies a dataset and benchmark for long-range sensing.
98. PARSEE-VAD: Efficient Training-Free Online Video Anomaly Detection via Proposition-Aware Reasoning and Streaming Evidence Escalation
Core Problem: Training-free online video anomaly detection (VAD) with frozen multimodal language models faces two coupled challenges: extracting reliable current-window semantics under causal and computational constraints, and maintaining temporal continuity without repeatedly transmitting high-dimensional history.
Key Innovation: Encoding history through text can compress visual evidence and introduce semantic bias, whereas retaining visual history expands multimodal context. These results support a current-first principle for streaming multimodal inference: resolve present semantics first, then use compact historical state only to repair residual continuity gaps.
99. Domain Generalization under Sampling Pattern Shifts in Irregular Time Series
Core Problem: Changing observation schedules create a deployment shift distinct from value noise.
Key Innovation: A domain-generalization study isolates sampling-pattern shifts in irregular time series.
100. BITS: Rethinking Fair and Comprehensive Evaluation for Irregular Time Series Forecasting
Core Problem: Forecast methods can be misranked when missingness and irregular sampling are evaluated inconsistently.
Key Innovation: BITS defines a broader evaluation of irregular time-series forecasting.
101. ReLoc: Rethinking Scene Coordinate Regression Architecture for Robust Outdoor LiDAR-based Localization
Core Problem: Scene Coordinate Regression (SCR) has recently emerged as a promising approach for LiDAR-based localization, achieving accurate localization without requiring an explicit 3D map.
Key Innovation: In this paper, we present ReLoc, a revamped SCR architecture that can effectively address these limitations. Experimental results on two large-scale outdoor datasets demonstrate that our approach achieves state-of-the-art accuracy over previous SCR-based methods while maintaining real-time inference performance.
102. Optimal Transport Dropout for Structured Predictive Uncertainty
Core Problem: Independent dropout masks can destroy correlated uncertainty structure.
Key Innovation: Optimal-transport dropout preserves structured predictive variation.
103. From Grey-Box to Green-Box: When can Physics-Informed Machine Learning Reduce Carbon Footprints in Structural Health Monitoring?
Core Problem: Machine learning plays an increasingly vital role in engineering, but the corresponding increase in compute time is not without environmental cost.
Key Innovation: Physics-informed machine learning or "grey-box" models have been developed to overcome some of the limitations of traditional black-box learners, utilising the physical insight that an engineer would have about the structure they are modelling and have shown promising results in the structural engineering field among many others. Although promising results, we cannot expect a silver bullet and the case studies demonstrate that a trade-off is needed between the increased complexity that comes from introducing physics into a machine learner, against the gain from reduced training data require...
104. TC-ADA: One-Shot Active Domain Adaptation for Semantic Segmentation
Core Problem: A single labelled target example must be used without overfitting.
Key Innovation: One-shot active adaptation selects and exploits a target observation for semantic segmentation.
105. How Synthetic Labels Improve Conformal Prediction: A Perspective on Conditional Coverage
Core Problem: Synthetic labels may improve marginal coverage while hiding groupwise failure.
Key Innovation: The paper analyzes when synthetic labels help conformal prediction and conditional coverage.
106. Correct then Forecast: Observer State-Space Models for Time Series Forecasting
Core Problem: Time series forecasting requires extrapolating the dynamics of an observed process beyond the last available measurement.
Key Innovation: Following a state-estimation perspective, we introduce Observer State-Space Models (OSSMs), a class of recurrent models that interprets the observed input time series as measurements of an underlying autonomous dynamical system. These results support a simple principle for recurrent forecasting: observations should correct the estimated latent state, rather than control the dynamics used to propagate it.
107. Revisiting Diffusion Fine-Tuning for Unsupervised Domain Adaptation
Core Problem: Diffusion-based unsupervised domain adaptation (UDA) improves cross-domain transfer by generating target-specific synthetic data for downstream adaptation.
Key Innovation: We propose MUSE (Multi-target UDA-oriented Synthesis with Efficient diffusion fine-tuning), a decoupled adaptation framework that separates source-supervised semantic adaptation from target-specific style adaptation. Experiments on standard UDA benchmarks show that MUSE achieves a stronger accuracy-efficiency trade-off than repeated per-target diffusion adaptation, reducing diffusion fine-tuning cost while improving average target-domain accuracy.
108. WhiteCon: Semi-Supervised Domain Adaptation Regression Through Whitening Transform and Dual Consistency
Core Problem: Domain adaptation is crucial for addressing distributional shifts that degrade model performance across domains.
Key Innovation: To address this gap, we propose semi-supervised domain adaptation regression through whitening transform and dual consistency (WhiteCon), which combines domain-specific whitening transform (DWT) and dual consistency regularization to enhance training stability and domain adaptation. Empirical evaluations on various benchmark datasets under SSDAR settings demonstrate that the proposed WhiteCon achieves state-of-the-art performance compared to existing methods, effectively addressing domain shifts in regression tasks.
109. Preference-Guided Adaptation for Open-Vocabulary Semantic Segmentation via Prompt Disagreement
Core Problem: Open-vocabulary semantic segmentation (OVSS) enables pixel-level prediction over arbitrary text-specified vocabularies and has shown strong generalization on common benchmarks.
Key Innovation: We propose a preference-guided adaptation framework that replaces dense mask supervision with binary preferences. Across extensive experiments on the MESS benchmark, the proposed method achieves consistent gains across diverse OVSS backbones without any pixel-level annotation, and remains effective under noisy preferences.
110. Instance-Adaptive Prompts as Context for Time-Series Foundation Models
Core Problem: Longer histories can improve time-series foundation models (TSFMs), but require substantially higher inference cost.
Key Innovation: We introduce PaCTS, which generates a small set of instance-adaptive latent prompts in the form of continuous embedding tokens conditioned on the visible context. Compared with weight-space adaptation methods, PaCTS achieves stronger improvements and better out-of-distribution generalization.
111. QiYao-M: Multimodal Time Series Foundation Model with Role-Aware Modeling of Endogenous and Exogenous Modalities
Core Problem: Existing multimodal time series foundation models (TSFMs) typically model heterogeneous modalities through largely shared mechanisms, overlooking the distinct forecasting roles of endogenous and exogenous modalities.
Key Innovation: In this work, we propose QiYao-M, a role-aware multimodal TSFM that models the two types of modalities separately. Extensive experiments across unimodal and multimodal benchmarks demonstrate strong forecasting performance in scenarios both with and without exogenous modalities.
112. Context-dependent time-series prediction via HyperReservoirs
Core Problem: Time series prediction is a common application of reservoir computing.
Key Innovation: Here, we propose a HyperReservoir as an extended model of reservoir computing especially designed for such cases. We find that the HyperReservoir achieves the lowest mean test error in all three tasks, and particularly outperforms conceptors on data that is sampled from the same attractor but at different time scales.
113. XMatch: Enhancing Covariate-Aware Time Series Forecasting through Tree-Structured Exogenous Matching
Core Problem: Future exogenous variables provide valuable information for forecasting endogenous time series.
Key Innovation: These associations motivate a strategy that matches future and historical exogenous patterns and uses the corresponding endogenous patterns to enhance forecasting. Extensive experiments on 12 real-world datasets demonstrate that XMatch outperforms state-of-the-art baselines.
114. Universality and Generalization of Causal Transformers Across Context Lengths
Core Problem: Long contexts are central to modern transformer systems, but most expressivity results choose a different network for each fixed sequence length.
Key Innovation: We study whether one masked transformer can approximate causal token-to-token maps uniformly over sequences of arbitrary length sampling a fixed normalized horizon. The result extends to the infinite-length mean-field limit, where tokens form continuous curves and masked attention becomes a causal time integral.
115. Spectral Super-Resolution using Spatial-Spectral Residual Operator Networks
Core Problem: Spectral super-resolution of multispectral satellite images can enable high temporal- and spatial-resolution hyperspectral satellite imagery at a modest cost, significantly increasing the applicability of hyperspectral remote sensing.
Key Innovation: In this study, the spectral super-resolution task is framed as an operator learning problem, and SSRON is proposed as a Deep Operator Network that effectively learns function-to-function mappings from downsampled spectra to continuous spectra. These results suggest the potential of SSRON and establishes operator learning as a promising direction for spectral super-resolution.
116. ReVA: A Scene-Centric Dataset Beyond Repetition for Remote Sensing Video Question Answering
Core Problem: Video question answering often ignores repeated scenes and temporal state.
Key Innovation: ReVA provides a scene-centric remote-sensing video benchmark beyond repetition.
117. Remote Sensing Sparse-View 3D Gaussian Splatting via Depth Image-Based Rendering
Core Problem: Sparse satellite or aerial views leave large geometric gaps.
Key Innovation: Depth-image rendering guides Gaussian reconstruction for sparse-view remote sensing.
118. A Unified Uncertainty Representation for Graph Neural Networks via Doubly-Spectral Stochastic Expansion
Core Problem: Node and spectral uncertainty are usually modeled separately.
Key Innovation: A doubly spectral stochastic expansion unifies uncertainty in graph neural networks.
119. Superquadric Primitive Decomposition of 3D point clouds via Geometric-Aware Inlier Refinement
Core Problem: Complex point clouds require compact geometric primitives.
Key Innovation: Superquadric decomposition uses geometric inlier refinement for 3D point clouds.
120. Uncertainty Quantification of Next Generation Reservoir Computing with Applications to Memory-Driven Dynamical Systems
Core Problem: Reservoir models need uncertainty estimates for memory-driven dynamics.
Key Innovation: Next-generation reservoir computing is paired with uncertainty quantification.
121. Certifying Interventional Agreement Among Observationally Equivalent Causal Models
Core Problem: Observationally equivalent causal models can imply different interventions.
Key Innovation: The work certifies when such models agree on interventional conclusions.
122. MixBench-TS: A Multivariate Time Series Forecasting Benchmark Where Channel Mixing Pays Off
Core Problem: Multivariate Time Series Forecasting (MTSF) models that mix information across channels assume that the past of one channel carries information about the future of another.
Key Innovation: Yet they are evaluated on a small fixed set of standard datasets whose cross-channel structure is rarely examined. Code and data are available at https://anonymous.4open.science/r/mixbench-ts-B027.
123. PINNMorph: Evolving Online Adaptation Policies for Physics-Informed Neural Networks
Core Problem: A fixed adaptation rule can fail as PDE regimes change.
Key Innovation: PINNMorph evolves online adaptation policies for physics-informed networks.
124. Domain Adaptation with Target Information via Doubly-Anchored Distributionally Robust Optimization
Core Problem: Target information can improve adaptation while also enabling overfitting.
Key Innovation: Doubly anchored distributionally robust optimization constrains domain adaptation.
125. Can Tabular Foundation Models Amortize Statistical Inference?
Core Problem: Tabular foundation models may reuse computation across repeated inference tasks.
Key Innovation: The paper tests whether tabular foundation models amortize statistical inference.
126. Sharp training-conditional coverage for conformal prediction under covariate shift
Core Problem: Training-conditional coverage can fail after covariate shift.
Key Innovation: Sharp conditional guarantees characterize conformal prediction under shifted covariates.
127. Sparsity by Default: The Theory and Practice of ARD in Gaussian Process Regression for Variable Selection
Core Problem: Dense covariate sets weaken interpretability and scaling.
Key Innovation: Automatic relevance determination is examined as a default variable-selection strategy for GPs.
128. The Statistical Cost of Causal Discovery with Feedback
Core Problem: Feedback raises the sample cost of distinguishing causal structure.
Key Innovation: The paper characterizes statistical limits for causal discovery with feedback.
129. GT-PSSM: Unified Probabilistic Framework for Stochastic Dynamics Modeling and Dependency Learning in Multivariate Time Series Anomaly Detection
Core Problem: Anomaly models need both stochastic dynamics and dependency learning.
Key Innovation: GT-PSSM combines probabilistic state modeling with multivariate anomaly detection.
130. Conformal Prediction and Conditional Coverage for Tabular Foundation Models
Core Problem: Foundation predictors need calibrated uncertainty, not only average accuracy.
Key Innovation: Conformal methods are evaluated with tabular foundation models and conditional-coverage diagnostics.
131. Multi-Task Learning of Conditional Mean Operators: applications to dynamical systems and uncertainty quantification
Core Problem: Multiple tasks may share transition structure but differ in conditional responses.
Key Innovation: Multi-task conditional mean operators support dynamics and uncertainty estimation.
132. OrientedFormer: An End-to-End Transformer-Based Oriented Object Detector in Remote Sensing Images
Core Problem: Axis-aligned detectors poorly represent rotated objects in aerial imagery.
Key Innovation: OrientedFormer performs end-to-end oriented detection in remote-sensing images.
133. DRAN: A Distribution and Relation Adaptive Network for Spatio-temporal Forecasting
Core Problem: Spatial relationships and data distributions vary during deployment.
Key Innovation: DRAN adapts both distributions and relations for spatiotemporal forecasting.
134. Dinomaly2: A Unified Framework for Unsupervised Image Anomaly Detection
Core Problem: Anomaly detectors often specialize to one texture or object regime.
Key Innovation: Dinomaly2 unifies unsupervised visual anomaly detection across benchmarks.
135. Sparse, self-organizing ensembles of local kernels detect rare statistical anomalies
Core Problem: Fixed ensembles can miss localized low-probability structure.
Key Innovation: Sparse self-organizing local kernels detect rare statistical anomalies.
136. PACGNet: Pyramidal Adaptive Cross-Gating Network for Multimodal Object Detection in Aerial Imagery
Core Problem: Object detection in aerial imagery is challenging because targets are often small, backgrounds are cluttered, and imaging conditions vary substantially.
Key Innovation: This paper presents PACGNet, a lightweight backbone-centric framework for multimodal aerial object detection. These results indicate a favorable accuracy-efficiency trade-off for lightweight multimodal aerial object detection.
137. Multivariate Time Series Forecasting needs Cross Variable Loss
Core Problem: Independent channel losses can ignore cross-variable forecast coherence.
Key Innovation: A cross-variable loss is proposed for multivariate time-series forecasting.
138. A Unified Risk View of Uncertainty: Posterior Risk for Disentanglement and Evaluation Beyond Proxies
Core Problem: Proxy uncertainty scores can disagree with decision risk.
Key Innovation: Posterior risk provides a common view for disentanglement and uncertainty evaluation.
139. Nonstationary cycle-damage accumulation and Rainflow-Miner correction for UHPC offshore wind turbine towers under threshold-exceedance response clusters
Core Problem: Conventional fatigue assessment of offshore wind turbine towers commonly relies on rainflow counting and Miner's rule, but this cycle-based approach may suppress temporal information when responses become nonstationary.
Key Innovation: This study investigates nonstationary cycle-damage accumulation in UHPC offshore wind turbine towers and proposes a window-wise correction to conventional Rainflow-Miner assessment. The framework requires independent calibration before field application.
140. The satellite-only gravity field model GOCO2025s
Core Problem: GOCO2025s is a global gravity field model represented by spherical harmonic coefficients up to degree and order 300, with constrained secular, annual, and semi-annual variations up to degree and order 200.
Key Innovation: It is derived from 23 years of satellite data, incorporating observations from the dedicated gravity field missions Gravity field and steady-state Ocean Circulation Explorer (GOCE), Gravity Recovery and Climate Experiment (GRACE), and GRACE Follow-On (GRACE-FO), as well as kinematic orbits of 18 low Earth orbit satellites and satellite laser ranging observations to 9 passive geodetic satellites. They comprise the static and time-dependent gravity field coefficients and their respective uncertainty information, all parts of the normal equation system of the static gravity field, and the grav...
141. Regional Groundwater Database for Arequipa and Surroundings in Southern Peru from Unstructured Sources Using an Integrated Optical Character Recognition and Large Language Model Workflow
Core Problem: Groundwater management in the arid Andes is constrained less by an absence of measurements than by their inaccessibility: hydraulic parameters measured over six decades remain in unpublished theses and consultancy reports that are not machine-readable and are indexed nowhere.
Key Innovation: We present a regional groundwater database for Arequipa, southern Peru, containing 4775 georeferenced records recovered from 3675 source documents together with the national monitoring portal, spanning 1966 to 2025. Data are archived on HydroShare under CC BY 4.0 (Venegas-Quiñones et al., 2026).
142. Automatic Cloud Segmentation and Cloud Cover Retrieval from Wide-Field Thermal Infrared Whole Sky Images
Core Problem: Thermal infrared (TIR) whole-sky imaging enables continuous cloud observations during both daytime and nighttime, making it an important tool for ground-based cloud monitoring.
Key Innovation: Moreover, wide-field imaging covers not only the zenith region but also the peripheral areas at large zenith angles, which are susceptible to the combined effects of strong atmospheric background emission, thermal radiation contamination from the surface and surrounding environment, and imaging geometric distortion, resulting in reduced cloud-background contrast and making cloud segmentation considerably more challenging than in conventional infrared sky imagery. These results prove that the proposed framework provides an effective and robust solution for long-term ground-based cloud monito...
143. TECIS-1 Cloud and Aerosol Detection Method and Comparison
Core Problem: The Terrestrial Ecosystem Carbon Inventory Satellite (TECIS-1) is a Chinese-developed satellite that employs both active and passive remote sensing technologies for forest carbon monitoring.
Key Innovation: In this study, we propose a complete workflow for cloud-aerosol detection and parameter retrieval using 532 nm and 1064 nm attenuated backscatter data from TECIS-1 Level 1B products, combined with GMAO MERRA-2 reanalysis data. These results indicate that TECIS-1 can be used for cloud and aerosol detection, and provide a methodological reference for data processing and application of Chinese spaceborne atmospheric lidar systems.
144. Active Acoustic Remote Sensing of Ocean Sound Speed Fields Along a Survey Track Using Inversion Constrained by Acoustic Propagation
Core Problem: The ocean sound speed profile (SSP) governs underwater acoustic propagation but remains difficult to observe continuously along a moving survey track using direct profiling instruments.
Key Innovation: This paper investigates differential acoustic propagation delays as remote observations that integrate propagation information along the acoustic paths for SSP retrieval along a survey track. These results support active acoustic remote sensing as a practical approach for rapid SSP retrieval along a survey track under the tested configuration using a single vessel.
145. Evaluation of vertical accuracy of five open-source DEM datasets in a mountainous city: A case study of Chongqing central urban area, China
Core Problem: Accurate digital elevation models (DEMs) are essential for terrain analysis, urban planning, hydrological modeling, and disaster risk assessment in mountainous cities.
Key Innovation: Taking the central urban area of Chongqing, China, as a representative mountainous city, this study evaluates the vertical accuracy of five freely available global DEMs, namely ALOS PALSAR, Copernicus DEM, NASADEM, AW3D30, and ASTER GDEM V3. These findings provide practical guidance for selecting suitable open-source DEMs in mountainous urban environments and highlight the need to consider terrain, land cover, and urban morphology when applying DEMs in complex cities.
146. Water isotopes-based investigation of hydrological processes in a climatologically sensitive glacierized catchment in the post-monsoon season
Core Problem: Recent studies in the headwaters of the Ganga River (Central Himalaya) have predicted an increase in post-monsoon temperature and wet precipitation with a decrease in snowfall, which could affect hydrology of glacierized catchment.
Key Innovation: This study employed stable water isotopes to investigate key hydrological processes governing movement of water in the Pindar River (a tributary of the Alaknanda River) basin and further estimated seasonal (post-monsoon) glacier-snowmelt contribution to the river using two-component mixing models. These results underscore the strong dependence of the Pindar River on glacier-snowmelt during post-monsoon, making the water resources of this central Himalayan region highly vulnerable to the impacts of ongoing climate change and accelerated glacier recession.
147. Elevation-dependent spring snowmelt rates and degree-day factors in the Irtysh River Basin, Northern Xinjiang
Core Problem: Under the combined influence of global climate change and regional climate variability, extreme events are occurring with increasing frequency, and the risk of snowmelt flood disasters is rising in Xinjiang, China.
Key Innovation: Compared with previous studies, the DDF values in the Irtysh River Basin are slightly lower but present more pronounced spatial heterogeneity, which can be attributed to regional differences in climatic and topographic conditions. Our results contribute to a deeper understanding of the rapid snowmelt process in spring and provide important quantitative references for snowmelt model parameterization scheme, so as to better assess and prevent snowmelt floods.
148. A Regional Intercomparison of Ice Sheet Mass Balance Estimates of the Northwest Greenland Ice Sheet
Core Problem: Changes in ice sheet mass manifest as changes in ice flow, ice thickness, and gravitational attraction.
Key Innovation: These parameters can be measured from space and used to track Greenland's contribution to sea level rise independently from each other. Finally, owing to the good regional agreement between our estimates, we construct a reconciled mass balance record, showing that the Northwest sector contributed 5.1 ± 0.2 mm to sea level rise between 1972 and 2024.
149. Propagation of Subduction Polarity Reversal Along Arc-Continent Convergent Margin: Insights From 3D Geodynamic Simulations
Core Problem: Subduction polarity reversal (SPR) driven by arc-continent collision stands as a crucial process to establish new subduction zones throughout the Phanerozoic.
Key Innovation: Despite its inherent 3D nature, investigations of SPR along extensive arc-continent collision margins have largely been confined to 2D models, providing limited perspectives on the collision-perpendicular propagation of SPR and the associated tectonic evolution. In contrast, simulations involving young or rheologically weak lithosphere result in slab tearing or slab break-off, thereby interrupting SPR propagation.
150. Divergent Influences of Oceanic Western Boundary Currents on Weak and Strong Tropical Cyclones
Core Problem: Tropical cyclones (TCs) in the Northern Hemisphere often cross oceanic western boundary currents (WBCs), yet how these warm currents modulate TC intensity remains unclear.
Key Innovation: We analyze TCs traversing the Kuroshio and Gulf Stream during 1993-2025 and find that weak TCs respond more favorably than strong TCs. Our findings highlight the growing role of WBC thermal conditions for coastal TC intensity prediction.
151. Onset of Thrusting at ca. 23 Ma in East Kunlun Mountains, Northern Tibet, Recorded by Syntectonic Strata
Core Problem: The timing of Cenozoic deformation in the East Kunlun Mountains, northern Tibet, is crucial for understanding plateau growth but remains disputed due to the scarcity of direct chronological constraints on specific tectonic structures.
Key Innovation: Here, we present a high-density-sampling magnetostratigraphic record from the Xiaohongshan section, interior of the East Kunlun Mountains, where growth strata directly constrain syndepositional thrust activity. Our results provide a direct temporal link between deep lithospheric processes and surface tectonics in northern Tibet.
152. Water Storage Capacity Dynamics in Drained Boreal Peatlands: Responses to Rewetting and Climate Change
Core Problem: Historical peatland drainage resulted in surface subsidence and reduced water storage capacity (WSC), a process that can be exacerbated by climate change.
Key Innovation: Results showed that rewetting significantly increased WSC under both present and future RCP 4.5 climates, with stronger effects where Sphagnum moss recovery may restore surface porosity. These results demonstrate that rewetting effectively restores the flood control function of drained boreal peatlands, strengthens their resilience under climate change, and supports ecosystem recovery and climate adaptation strategies.
153. Quantifying Land-Sea Connectivity Through Remotely Sensed River Plumes
Core Problem: Coastal ecosystems are highly influenced by riverine freshwater inputs, but the nature of these influences is dynamic and hard to characterize.
Key Innovation: Here we present a method that leverages remote sensing data of river plumes to quantify land-sea connectivity. We anticipate applications of the proposed method will lead to improved understanding of linkages between the terrestrial and marine environment, which is needed to guide coastal management, conservation, and restoration.
154. Quantifying Dead-End Impacts on Radon Signals in Fractured Rocks
Core Problem: Understanding radon transport in fractured rock is fundamental for interpreting geochemical signals associated with subsurface deformation and fluid transport.
Key Innovation: The role of dead-end fractures (DEFs), one key geometric feature of fractured rocks, remains poorly understood. This study advances our mechanistic understanding of radon nuclide transport in fractured rocks.
155. Thickness-Dependent Thermodynamic Responses of Arctic Sea Ice to Extratropical Cyclones Based on MOSAiC Observations
Core Problem: Extratropical cyclones strongly influence Arctic sea ice thermodynamics through episodic heat and moisture transport, yet their thickness-dependent impacts remain poorly quantified.
Key Innovation: Using comprehensive observations from the Multidisciplinary Drifting Observatory for the Study of Arctic Climate (MOSAiC), we investigate how sea ice thickness (SIT) regulates thermodynamic responses during cyclones. Among them, exhibits the strongest sensitivity to SIT, followed by, whereas and show weaker dependence because synoptic signals are progressively damped during downward propagation.
156. Investigation of Ultraclose Jet Grouting on Existing Cable Tunnel: Field Monitoring and Analytical Insights
Core Problem: Ultraclose jet grouting (JG) inevitably induces significant disturbances to adjacent infrastructure, such as tunnels.
Key Innovation: Then, an analytical model informed by the monitoring results was proposed to investigate the multimodal deformation behavior and force redistribution in the cable tunnel. The key findings are as follows: (1) In the later stages of double-fluid JG, clogging or discontinuous operations easily cause pronounced cyclic uplift-settlement deformation of the tunnel.
157. A hybrid structural response cross-correlation functions based damage identification strategy for offshore platforms
Core Problem: Title-level focus: Publication date: 15 November 2026 Source: Ocean Engineering, Volume 368, Part 3 Author(s): Xiaojuan Wang, Fan Yang, Hongyuan Zhou, Xiangyong Lan, Jian Zhang, Nannan Shi, Lihui Wang
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
158. In-situ shear-wave sound speed and attenuation in sandy seabed sediment: frequency dependence from 500 Hz to 3 kHz
Core Problem: Title-level focus: Publication date: 15 November 2026 Source: Ocean Engineering, Volume 368, Part 3 Author(s): Yuhao Zhang, Jingqiang Wang, Guanbao Li, Qingjie Zhou, Xiangmei Meng, Qingfeng Hua, Guangming Kan, Chenguang Liu
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
159. Integrating anthropogenic hydrodynamics and GOCI satellite observations for water quality retrieval in highly engineered coastal systems: A case study of Saemangeum
Core Problem: Title-level focus: Publication date: 15 December 2026 Source: Remote Sensing of Environment, Volume 347 Author(s): Yejin Lee, Dongjin Kim, Suhwan Kim, Dohee Han, Hyun-su Kim, Jonghan Ko, Su-mi Kim, Kyeong-sang Lee, Jong-Min Yeom
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
160. Mechanical response of large-diameter reinforced concrete curved pipe jacking: Field full-scale test and simulation
Core Problem: Title-level focus: Publication date: January 2027 Source: Tunnelling and Underground Space Technology, Volume 179, Part 2 Author(s): Rudong Wu, Kaixin Liu, Cong Zeng, Hui Xie, Jingliang Ye
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
161. Effect of relative humidity on the desiccation behavior of bentonite buffer blocks for geological repositories
Core Problem: Title-level focus: Publication date: March 2027 Source: Computers and Geotechnics, Volume 203, Part 1 Author(s): Yuan Feng, Jongwan Eun, Seunghee Kim, Yong-Rak Kim
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
162. Uncertainty in physical modeling of vertical breakwaters induced by wave phase variability
Core Problem: Title-level focus: The study addresses the focus identified by its title: Uncertainty in physical modeling of vertical breakwaters induced by wave phase variability.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
163. DEViCo: Discriminative Evidence View Correspondence Learning for Drone-Satellite Cross-View Geo-Localization
Core Problem: Title-level focus: The study addresses the focus identified by its title: DEViCo: Discriminative Evidence View Correspondence Learning for Drone-Satellite Cross-View Geo-Localization.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
164. FMC-DETR: Frequency-Decoupled Multi-Domain Coordination for Aerial-View Object Detection
Core Problem: Title-level focus: The study addresses the focus identified by its title: FMC-DETR: Frequency-Decoupled Multi-Domain Coordination for Aerial-View Object Detection.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
165. Unveiling profile distribution of soil organic carbon in China’s largest freshwater wetland using a novel 3D INLA-SPDE model by integrating remote sensing and ground field survey data
Core Problem: Title-level focus: The study addresses the focus identified by its title: Unveiling profile distribution of soil organic carbon in China’s largest freshwater wetland using a novel 3D INLA-SPDE model by integrating remote sensing and ground field survey data.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
166. First Demonstration of an Onboard Polarimetric Calibration and Ice-Cloud-Based Validation Framework for the Particulate Observing Scanning Polarimeter (POSP)
Core Problem: Title-level focus: The study addresses the focus identified by its title: First Demonstration of an Onboard Polarimetric Calibration and Ice-Cloud-Based Validation Framework for the Particulate Observing Scanning Polarimeter (POSP).
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
167. Geo-Sparse Reconstruction Transformer for Remote Sensing Image Super-Resolution
Core Problem: Title-level focus: The study addresses the focus identified by its title: Geo-Sparse Reconstruction Transformer for Remote Sensing Image Super-Resolution.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
168. 3-D MCSEM Forward Modeling with Mixed-Order Spectral-Element Method Based on Nonconformal Meshes
Core Problem: Title-level focus: The study addresses the focus identified by its title: 3-D MCSEM Forward Modeling with Mixed-Order Spectral-Element Method Based on Nonconformal Meshes.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
169. HRSF: A Hierarchical Residual Distribution Framework for Remote Sensing Image Fusion under Spatiotemporal Constraints
Core Problem: Title-level focus: The study addresses the focus identified by its title: HRSF: A Hierarchical Residual Distribution Framework for Remote Sensing Image Fusion under Spatiotemporal Constraints.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
170. Seismic Petrophysical and Lithofacies Joint Inversion with Gaussian Mixture Model and Markov Random Field
Core Problem: Title-level focus: The study addresses the focus identified by its title: Seismic Petrophysical and Lithofacies Joint Inversion with Gaussian Mixture Model and Markov Random Field.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
171. SCD-MoE: Adaptive Expert Routing for Efficient Remote Sensing Semantic Change Detection
Core Problem: Title-level focus: The study addresses the focus identified by its title: SCD-MoE: Adaptive Expert Routing for Efficient Remote Sensing Semantic Change Detection.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
172. Radar-Referenced Prior Network for Infrared Iceberg Detection
Core Problem: Title-level focus: The study addresses the focus identified by its title: Radar-Referenced Prior Network for Infrared Iceberg Detection.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
173. Intelligent Pre-stack Common Offset Gather Denoising with Post-stack Structure Guidance
Core Problem: Title-level focus: The study addresses the focus identified by its title: Intelligent Pre-stack Common Offset Gather Denoising with Post-stack Structure Guidance.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
174. End-to-End Evolving Cross-Modal Graph Network for Unsupervised Multi-Modal Remote Sensing Change Detection
Core Problem: Title-level focus: The study addresses the focus identified by its title: End-to-End Evolving Cross-Modal Graph Network for Unsupervised Multi-Modal Remote Sensing Change Detection.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
175. Geo-ARD: A Hybrid ARDB-Guided Residual Diffusion Model for Geospatial Wind Field Super-Resolution
Core Problem: Title-level focus: The study addresses the focus identified by its title: Geo-ARD: A Hybrid ARDB-Guided Residual Diffusion Model for Geospatial Wind Field Super-Resolution.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
176. Model-Driven GPR Inversion Network With Surrogate Forward Solver
Core Problem: Title-level focus: The study addresses the focus identified by its title: Model-Driven GPR Inversion Network With Surrogate Forward Solver.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
177. Validating high resolution evapotranspiration estimates with eddy covariance flux measurements: a disaggregation approach to resolve spatial scale mismatch
Core Problem: Title-level focus: The study addresses the focus identified by its title: Validating high resolution evapotranspiration estimates with eddy covariance flux measurements: a disaggregation approach to resolve spatial scale mismatch.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
178. Urban LiDAR Point Cloud Segmentation Using Multi-Level Stacked Learning With Counterfactual Guided Minority Class Enhancement
Core Problem: Title-level focus: The study addresses the focus identified by its title: Urban LiDAR Point Cloud Segmentation Using Multi-Level Stacked Learning With Counterfactual Guided Minority Class Enhancement.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
179. Hydrogeochemical evolution of a glacier-fed Himalayan mountain system: Upper Ganga Basin
Core Problem: Title-level focus: The study addresses the focus identified by its title: Hydrogeochemical evolution of a glacier-fed Himalayan mountain system: Upper Ganga Basin.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
180. A positivity-constrained physics-informed Kolmogorov-Arnold network for stability-oriented source-intensity inversion in nuclear-contaminant transport
Core Problem: Title-level focus: The study addresses the focus identified by its title: A positivity-constrained physics-informed Kolmogorov–Arnold network for stability-oriented source-intensity inversion in nuclear-contaminant transport.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
181. The volume balance of the Andean convergent margin and its control on coupling between surface processes and tectonics
Core Problem: Title-level focus: The study addresses the focus identified by its title: The volume balance of the Andean convergent margin and its control on coupling between surface processes and tectonics.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
182. Denudation of the Australian continent rates, controversies and landscape evolution
Core Problem: Title-level focus: The study addresses the focus identified by its title: Denudation of the Australian continent rates, controversies and landscape evolution.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
183. Remote sensing video super-resolution via various real-world degradation modeling
Core Problem: Title-level focus: The study addresses the focus identified by its title: Remote sensing video super-resolution via various real-world degradation modeling.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
184. Fine-scale annual mapping and dynamic monitoring of shrub encroachment in alpine grasslands by integrating Sentinel-2 phenological features with Sentinel-1 structural information
Core Problem: Title-level focus: The study addresses the focus identified by its title: Fine-scale annual mapping and dynamic monitoring of shrub encroachment in alpine grasslands by integrating Sentinel-2 phenological features with Sentinel-1 structural information.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
185. Recirculation tracer tests in long uncased boreholes: a simple way to identify hydraulic conductivity
Core Problem: Title-level focus: The study addresses the focus identified by its title: Recirculation tracer tests in long uncased boreholes: a simple way to identify hydraulic conductivity.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
186. The impact of numerical error on hydrological model performance
Core Problem: Title-level focus: The study addresses the focus identified by its title: The impact of numerical error on hydrological model performance.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
187. Probabilistic sensor placement in water distribution networks: a Dynamic Residual Masking framework using game-theoretic pseudo-Jacobians
Core Problem: Title-level focus: The study addresses the focus identified by its title: Probabilistic sensor placement in water distribution networks: a Dynamic Residual Masking framework using game-theoretic pseudo-Jacobians.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
188. Scaling properties and classification of urban drainage networks in an Asian megacity
Core Problem: Title-level focus: The study addresses the focus identified by its title: Scaling properties and classification of urban drainage networks in an Asian megacity.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
189. Deep learning meets spectral analysis: reproducing long-term persistence in synthetic streamflow
Core Problem: Title-level focus: The study addresses the focus identified by its title: Deep learning meets spectral analysis: reproducing long-term persistence in synthetic streamflow.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
190. Coupling water balance and vegetation carbon dynamics in a parsimonious global hydrologic model
Core Problem: Title-level focus: The study addresses the focus identified by its title: Coupling water balance and vegetation carbon dynamics in a parsimonious global hydrologic model.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
191. Finite element modelling of pile penetration in sand considering the effect of fabric anisotropy
Core Problem: Title-level focus: The study addresses the focus identified by its title: Finite element modelling of pile penetration in sand considering the effect of fabric anisotropy.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
192. High-strain-rate dynamic response and energy dissipation of loess under passive confinement: an SHPB investigation
Core Problem: Title-level focus: The study addresses the focus identified by its title: High-strain-rate dynamic response and energy dissipation of loess under passive confinement: an SHPB investigation.
Key Innovation: Title-signalled contribution: The available evidence identifies a concrete method or application, but does not support a stronger claim about validation or generalization. Methods, results and validation could not be assessed from a reliable abstract.
193. Enhancing Foundation Models for Imbalanced SAR Ship Classification via Targeted Oversampling
Core Problem: Remote-sensing foundation models offer strong representations for SAR imagery, but their behavior under severe long-tail class imbalance is still not well characterized.
Key Innovation: We benchmark DOFA and SAR-JEPA on the imbalanced OpenSARShip dataset and compare them with ImageNet-pretrained baselines under a fixed, training-efficient protocol that keeps the backbone frozen. We also report class-wise behavior, showing that aggregate improvements can coexist with persistent failures on specific rare classes.
194. Cross-Dataset Transfer and Unknown-Class Detection in Imbalanced SAR Ship Classification
Core Problem: Ship classification from Synthetic Aperture Radar (SAR) imagery is a critical computer vision task, yet the robustness of models under deployment shifts remains unclear.
Key Innovation: To address this, we evaluate six pretrained models on two SAR ship datasets in three settings: in-domain classification, cross-dataset transfer, and unknown-class detection. These findings show that cross-dataset generalization in SAR remains limited and that task-specific uncertainty scores are often more informative than MC-dropout variance for held-out-class detection, although their relative ranking depends on the dataset and held-out class.
195. Active Causal Discovery Benchmark: Evaluating LLM Agents Under Budgeted Interventions
Core Problem: We introduce the Active Causal Discovery Benchmark (ACDB), an SCM-grounded environment for evaluating whether LLM agents recover causal graph structure from observations and budget-constrained hard interventions.
Key Innovation: The most informative diagnostic is the precision-recall decomposition: PC under-commits with high precision, LLMs over-commit with lower precision, and statistical-tool access often increases abstention rather than useful intervention. The current results should therefore be read as a benchmark audit and calibration report, not as evidence that current LLMs solve active causal discovery.
196. Does Joint-Embedding Predictive Architecture Pretraining Help Time Series Forecasting?
Core Problem: Joint-embedding predictive architectures (JEPA) have emerged as a promising self-supervised pretraining paradigm for time series, learning representations by predicting target embeddings in latent space rather than reconstructing raw signals.
Key Innovation: Yet evidence on their benefits remains mixed, and most studies test only a single backbone or a narrow set of architectures, leaving unclear whether JEPA pretraining is a reliable improvement or one that depends heavily on the downstream model. We find that the benefit of this instantiation varies sharply across backbones, producing consistent gains for some architectures and consistent degradation for others, even on the same dataset.
197. SWT: Self-Supervised Video Object Segmentation via Sliding, Wavelet and Transportation
Core Problem: Video Object Segmentation (VOS) aims to accurately segment target objects from consecutive video frames and track the changes of the objects in each frame of the video.
Key Innovation: Based on this observation, in this paper, we propose self-supervised VOS with Sliding window, Wavelet transform and optimal Transport (SWT), a self-supervised VOS framework entirely trained on static dataset using contrastive learning. SWT only requires training on the COCO dataset once and achieves excellent results on five VOS datasets as well as an additional body part propagation dataset.
198. USAI-Quant: A Quantitative Reasoning Benchmark for Vision-Language Models in Built Environments
Core Problem: Large vision-language models (VLMs) have emerged as a powerful paradigm for urban and spatial AI.
Key Innovation: To address this gap, we develop Quantitative Urban and Spatial AI benchmark (USAI-Quant), the first benchmark designed to quantitatively evaluate VLM's reasoning capabilities on built environment metrics via remote sensing imagery. Our results reveal that current state-of-the-art models consistently fall short on numeric reasoning tasks.
199. The Selection Rule Decides the Winner: A Pre-Registered Audit of Open-Set Graph Anomaly Detection
Core Problem: Benchmark conclusions can depend on how open-set scores are selected.
Key Innovation: A preregistered audit shows that the selection rule changes rankings in graph anomaly detection.
200. From HL to H+L-1 Parameters: A Hankel-Toeplitz Forecaster for Long-Term Time Series Forecasting
Core Problem: Linear forecasters have shown competitive accuracy against Transformer-based models in long-term time series forecasting.
Key Innovation: HTF uses H+L-1 trainable coefficients while allowing a full-rank forecasting matrix.
201. CRF Loss is How Networks Should Learn Boundaries in Weakly Supervised Segmentation
Core Problem: Weakly Supervised Semantic Segmentation (WSSS) learns pixel-level predictions from image-level tags.
Key Innovation: Inspired by CRF potentials, we propose a framework that disentangles soft pseudo-labels as unary supervision and binary edge maps as pairwise supervision. Recent work focuses on improving coarse CAMs extracted from large vision-language models (commonly CLIP), but does little to improve their accuracy along segment boundaries.
202. HyperDAM: Hyperspectral Distractor-Aware Memory with Amodal Expansion for SAM 3 Tracking
Core Problem: Hyperspectral video provides material cues that can disambiguate targets with similar false-color appearance, yet foundation-model trackers update memory primarily from spatial and appearance evidence.
Key Innovation: We present HyperDAM, a DAM4SAM3-based hyperspectral tracker with three principal contributions. The final system ranked second in HOTC 2026, achieving 68.0093% AUC and 87.7703% DP@20 in the organizer's private evaluation.
203. MASCIT: A Mask-Aware State Space Classifier for Naturally Irregular Time Series
Core Problem: Naturally irregular time series combine asynchronous observations, missing values, unequal lengths, and nonuniform sampling, while dense adapters can discard temporal structure.
Key Innovation: We propose a mask-aware state space classifier for irregular time series (MASCIT), which supplies observation masks to the encoder and excludes invalid steps from gated temporal aggregation. These results support selective state space models as effective, executable backbones for naturally irregular time series classification.
204. Memory- and Bandwidth-Efficient SPAD-LiDAR Ranging via Coarse-to-Fine Spline Sketching
Core Problem: This work presents an optimized compression framework for direct time-of-flight (dToF) light detection and ranging (LiDAR), using an accurate, compact timestamp-encoding strategy to address the high data-rate bottleneck from single-photon timing information in single-photon avalanche diode (SPAD) arrays.
Key Innovation: We propose a hardware-friendly timestamp-to-depth framework in which a sparse single-photon encoding strategy, namely coarse-to-fine spline sketches (CFSS), projects photon timestamps into fine-grained, low-dimensional representations, called sketch values. Compared with the conventional sketched-LiDAR framework, the proposed method retains the same compression ratio in the single-peak case, ranging from hundreds× to thousands× depending on the accuracy-complexity trade-off, while improving depth-estimation accuracy.
205. Training-Free Uncertainty Estimation for Embedding Models
Core Problem: Embedding similarity lacks an uncertainty estimate.
Key Innovation: A training-free method estimates uncertainty for embedding models.